Abstract

MATERIALS AND METHODS

RESULTS

Study 1: Long-Term Effects of KD Feeding

KD feeding results in weight loss in wt but not ob/ob mice.

Consistent with our previous findings, after 60 days of study, KD-fed wt mice weighed significantly less than chow-fed controls. Age-matched ob/ob mice initially weighed an average of 21.6 g more than wt controls. In contrast to the wt groups, KD feeding did not lead to a significant change in weight of ob/ob mice when fed KD (Fig. 1A). Chow-fed ob/ob mice had higher caloric intake than chow-fed wt controls (Fig. 1B). However, the core temperature in both chow- and KD-fed ob/ob groups were significantly lower than wt mice (Fig. 1C). Furthermore, ob/ob groups had a significantly lower weight-adjusted o₂ (Fig. 1D) although comparison of heat production over a 72-h period of indirect calorimetry revealed no significant difference between chow- and KD-fed ob/ob mice (Fig. 1E). The respiratory exchange ratio approached 0.7 in KD-fed wt and ob/ob mice (Fig. 1F), consistent with a predominantly lipid fuel-based metabolism during KD feeding.

Open in a separate window

Fig. 1.

Effects of ketogenic diet (KD) feeding on weight and energy expenditure. A: 60 days of KD feeding led to sustained weight loss in wild-type (wt) but not ob/ob mice. Filled circles, chow-fed wt; open circles, KD-fed wt; filled squares, chow-fed ob/ob mice; open squares, KD-fed ob/ob mice. B: caloric intake was greater in chow-fed ob/ob mice compared with wt controls. Filled bars, chow fed; open bars, KD fed. *P < 0.001 wt vs. ob/ob. C: core temperature was lower in ob/ob mice but was not different in chow (filled bars) and KD (open bars)-fed groups. *P < 0.001 wt vs. ob/ob. D: indirect calorimety revealed increased o₂ in KD-fed wt groups that was apparent in the dark phase (filled bars) or light phase (open bars). This effect was absent in ob/ob mouse groups. E: heat production during 48-h period of indirect calorimetry revealed no difference between chow-fed and KD-fed wt mice or chow-fed and KD-fed ob/ob mice. F: respiratory exchange ratio (RER) was markedly decreased in KD-fed groups in both the dark phase and light phase (*P < 0.001 chow vs. KD).

KD improves glucose metabolism in ob/ob mice.

Following 50 days of KD feeding, there was significant improvement in insulin sensitivity of ob/ob mice despite the absence of weight change. Fasting glucose levels fell from 251 ± 38 mg/dl in chow-fed ob/ob mice to 110 ± 17 mg/dl in KD-fed ob/ob mice (Fig. 2A). Similarly, glucose excursions during the intraperitoneal glucose tolerance test were attenuated by KD feeding of ob/ob mice (Fig. 2A). KD-fed groups exhibited reduced fasted insulin levels in both wt (0.87 ± 0.13 ng/ml chow vs. 0.23 ± 0.09 ng/ml KD) and ob/ob (4.89 ± 0.37 ng/ml chow vs. 2.83 ± 0.36 ng/ml) mouse groups (Fig. 2B). Furthermore, the glucose-induced increment in circulating insulin in response to glucose challenge was restored in KD-fed ob/ob mice (Fig. 2B). As previously shown, chow-fed ob/ob mice were remarkably resistant to insulin treatment, with a maximal decrease in circulating glucose to 85 ± 7% basal levels. In contrast, glucose levels fell to 55 ± 6% basal levels following insulin administration in KD-fed ob/ob mice (Fig. 2C). Improved glycemia in ob/ob mice was seen in the context of a 75% fall of insulin levels. Calculation of QUICKI, a surrogate marker of insulin sensitivity (19–23), revealed that this metric was increased from 0.31 ± 0.01 in chow-fed to 0.41 ± 0.03 in KD-fed wt and from 0.23 ± 0.00 in chow-fed to 0.27 ± 0.01 in KD-fed ob/ob mice (Fig. 2D).

Open in a separate window

Fig. 2.

Effects of KD feeding on glucose metabolism. A: KD feeding normalized fasting glucose levels and reduced glucose excursions in ob/ob mice. Open circles, chow-fed wt; filled circles, KD-fed wt; open squares, chow-fed ob/ob mice; filled squares, KD. *P < 0.001 chow- vs. KD-fed ob/ob. B: KD feeding reduced fasting insulin levels. Filled bars, fasted; open bars, 45 min insulin. +P < 0.05 chow- vs. KD-fed wt mice. *P < 0.001 chow- vs. KD-fed ob/ob mice. Increment in circulating insulin in response to glucose challenge returned in KD-fed ob/ob mice. #P < 0.01 wt chow-fed basal vs. 45 min and ob/ob KD-fed basal vs. 45 min. C: insulin tolerance test revealed increased sensitivity in KD-fed ob/ob mice. Filled circles, chow-fed wt; open circles, KD-fed wt; filled squares, chow-fed ob/ob mice; open squares, KD. *P < 0.001 chow- vs. KD-fed ob/ob mice. D: quantitative insulin sensitivity check index (QUICKI) metric was greatly improved by KD feeding of ob/ob mice. Filled bars, chow; open bars, KD. #P < 0.01 chow- vs. KD-fed wt mice. *P < 0.001 chow- vs. KD-fed ob/ob mice.

KD feeding alters lipid profile.

KD feeding led to a 22% reduction in fed serum triglycerides in wt mice. Although there was a 13% trend toward lower triglycerides in KD-fed ob/ob mice, this effect did not reach statistical significance (Table 1). NEFA were elevated in response to KD feeding in wt mice; no increment was seen in the already elevated NEFA levels in KD-fed ob/ob mice. Of note, KD-fed wt mice became overtly ketotic, whereas there was only a small, albeit significant, change in the circulating β-hydroxybutyrate concentration in KD-fed ob/ob mice. There was a 48% increase in circulating cholesterol in KD-fed wt mice; however, the 7% trend toward increased circulating cholesterol in KD-fed ob/ob mice did not reach statistical significance (Table 1). Hepatic triglyceride was increased in KD-fed wt mice compared with chow-fed wt controls. However, there was no additional accumulation in the already fatty livers of KD-fed ob/ob mice (Table 1).

Table 1.

Effects of KD feeding on wt and ob/ob mice

	Wild Type		ob/ob
	Chow	KD	Chow	KD
Insulin, ng/ml	2.16±0.48*†‡	1.32±0.08†‡	19.6±5.0*‡§	5.4±1.3*†§
Glucagon, pmol/l	15.1±0.9†‡	17.7±2.5†	38.6±9.1*‡§	9.7±0.7†§
NEFA, meq/l	0.51±0.02†‡	0.61±0.04§	0.64±0.06§	0.66±0.07§
Glucose, mg/dl	248±9*†‡	155±16†‡§	296±12*‡§	205±17*†§
Triglyceride, mg/dl	87.2±6.7*‡	68.2±2.5†‡§	135±28*	117±14*§
β-Hydroxybutyrate, mmol/l	0.44±0.03*‡	2.48±0.3*‡†	0.50±0.04*‡	0.75±0.09*†§
Total cholesterol, mg/dl	94.0±3.3*†‡	139±4.4†‡§	174±9.3*§	187±5.0*§
Liver triglyceride, mg/g	3.50±0.10*†‡	8.94±0.28§	8.81±0.72§	8.10±0.82§

Open in a separate window

Data are presented as means ± SE; n = 5–8 mice. NEFA, nonesterified fatty acid. Wild-type (wt) or ob/ob mice fed ketogenic diet (KD) for 60 days. Mice were compared with age- and initial weight-matched chow-fed control groups. Circulating hormones and metabolites were measured in blood collected at the time of death (9:00 A.M.) while mice were allowed access to food and water ad libitum. Liver triglyceride was calculated from assay of extract from liver flash-frozen at the time of death.

P < 0.05 compared with wt chow (§)-,

wt KD (*)-,

ob/ob chow (†)-,

and ob/ob KD (‡)-fed groups.

KD feeding alters hepatic gene expression in wt and ob/ob mice.

Analysis of metabolic enzyme gene expression profiles in KD-fed ob/ob mice showed changes previously noted in wt mice consuming KD. Increases in hepatic expression of gluconeogenic enzymes glyceraldehyde-3-phosphate dehydrogenase and the rat- limiting phosphoenolpyruvate carboxykinase failed to reach statistical significance in ob/ob mice over wt controls (data not shown). As expected, expression of enzymes of lipid synthesis FAS and SCD1 were grossly elevated by 6.3- and 8.5-fold, respectively, in chow-fed ob/ob mice compared with wt controls (Fig. 3, A and B). Notably, feeding KD led to a marked suppression of FAS and SCD1 to 12.6 and 0.5% of chow-fed values, respectively, in ob/ob mice despite the absence of weight loss. Expression of genes required for fatty acid transport and oxidation also showed marked changes; the fatty acid translocase CD36 was upregulated in chow-fed ob/ob mice compared with wt controls (P < 0.005), and there was further induction of CD36 in KD-fed ob/ob mice (Fig. 3C). There was no difference in expression of enzymes of β-oxidation, long-chain acyl-coenzyme A dehydrogenase and 3-hydroxyacyl-coenzyme A dehydrogenase (HADH), between chow-fed wt or chow-fed ob/ob mice, and both were induced approximately twofold by KD feeding in mice of both genotype (Fig. 3, D and E). Notably, another hepatic PPARα target, uncoupling protein 2 (UCP2), was also upregulated by KD feeding by 12-fold in wt and 3.6-fold in ob/ob mice (Fig. 3F). Genes of ketosis, including 3-hydroxy-3-methylglutaryl-coenzyme A synthase 2 (HMGCS2) and 3-hydroxybutyrate dehydrogenase (HBDH) were also both induced approximately twofold by KD feeding from similar baseline levels in wt and ob/ob mice (Fig. 3, G and H).

Open in a separate window

Fig. 3.

Effects of KD feeding on expression of enzymes of hepatic metabolism. A and B: enzymes of lipid synthesis [fatty acid synthase (FAS) and stearoyl-coenzyme A desaturase 1 (SCD1)]. C–E: genes required for fatty acid transport and oxidation [fatty acid translocase (CD36) and long-chain acyl-coenzyme A dehydrogenase (ACADL) and 3-hydroxyacyl-coenzyme A dehydrogenase (HADH)]. F: peroxisome proliferator-activated receptor (PPAR)α target gene uncoupling protein-2 (UCP2). G and H: genes of ketosis [3-hydroxy-3-methylglutaryl-coenzyme A synthase 2 (HMGCS2) and 3-hydroxybutyrate dehydrogenase (HBDH)]. Filled bars, chow fed; open bars, KD fed. +P < 0.05, #P < 0.01, and ^*P < 0.001 chow vs. KD fed.

Hepatic FGF21 is induced by KD feeding in wt but not ob/ob mice.

Because hepatic FGF21 plays a vital role in mediating the effects of KD, we examined expression in ob/ob mice. Hepatic FGF21 mRNA levels were induced 10-fold in chow-fed ob/ob mice compared with wt controls. There was no further induction above already elevated levels by KD feeding of ob/ob mice, in contrast to wt mice in which KD feeding led to marked induction of FGF21 (Fig. 4A). Consistent with elevated activity of peroxisome proliferator-activator receptor (PPAR)α, expression of PPARα was itself increased twofold in chow-fed ob/ob mice. Expression of PPARγ was also elevated in chow-fed ob/ob mice, and, in common with PPARα, there was no significant difference between chow- and KD-fed groups of ob/ob mice (Fig. 4, B and C). Similarly, liver expression of PPARγ coactivator-1β, a key regulator of hepatic lipid metabolism, was upregulated in ob/ob mice but showed no interaction with KD feeding (Fig. 4D). In contrast, expression of sterol-regulatory element-binding protein-1c (SREBP1c) was downregulated by KD feeding in wt mice, a suppression that was blunted in ob/ob mice (Fig. 4E). Further analysis of expression of FGF21 in adipose tissue revealed increased expression in chow-fed ob/ob mice but no changes with KD feeding (Table 2).

Open in a separate window

Fig. 4.

Effects of KD feeding on hepatic regulatory gene expression. A: fibroblast growth factor 21 (FGF21). B: PPARα. C: PPARγ. D: PPARγ coactivator-1α (PGC1α). E: PPARγ coactivator-1β (PGC1β). F: sterol-regulatory element-binding protein-1c (SREBP1c). Filled bars, chow fed; open bars, KD fed. +P < 0.05, #P < 0.01, and ^*P < 0.001 wt vs. ob/ob mice.

Table 2.

Effects of KD feeding on tissue inflammatory markers in wt and ob/ob mice

	Wild Type		ob/ob
	Chow	KD	Chow	KD
Liver
IL-1	1.00±0.14*†‡	1.84±0.28§	1.66±0.27§	1.96±0.39§
IL-6	1.00±0.22	1.98±0.54	1.28±0.23	1.11±0.19
TNF-α	1.00±0.44*†‡	2.82±0.44§	3.5±0.75§	3.74±0.78§
SOCS3	1.00±0.41‡	1.33±0.35	1.26±0.38	2.74±0.62§
Ccl2	1.00±0.30*†‡	8.59±0.55§	8.55±0.81§	8.17±0.83§
Perigonadal adipose tissue
FGF21	1.00±0.21†‡	1.69±0.48†‡	4.90±1.00*§	5.97±0.37*§
IL-1	1.00±0.12†	1.45±0.19	1.62±0.26§	1.31±0.16
IL-6	1.00±0.12*†‡	0.65±0.11†‡§	2.45±0.51*§	2.14±0.34*§
TNF-α	1.00±0.18	0.81±0.09	4.00±0.87*§	4.95±0.11*§
SOCS3	1.00±0.19†‡	0.82±0.14†‡	2.12±0.50*§	2.04±0.28*§
Ccl2/MCP1	1.00±0.15†‡	0.69±0.11†‡	10.00±2.41*§	7.73±0.37*§
Serpine1/PAI1	1.00±0.16*†‡	2.23±0.57†‡§	4.07±0.88*§	4.81±0.35*§

Open in a separate window

Data are presented as means ± SE; n = 5–8 mice. IL, interleukin; TNF, tumor necrosis factor; SOCS, suppressor of cytokine signaling; Ccl, chemokine (C-C motif) ligand; FGF, fibroblast growth factor; MCP, monocyte chemotactic protein; PAI, plasminogen activator inhibitor. Wild-type or ob/ob mice were fed KD for 60 days. Mice were compared with age- and initial weight-matched chow-fed control groups. Liver and perigonadal adipose tissues were harvested and flash-frozen at the time of death. Assays were performed by quantitative PCR using Taqman gene expression assays and master mix.

P < 0.05 compared with wt chow (§)-,

wt KD (*)-,

ob/ob chow (†)-,

and ob/ob KD (‡)-fed groups. KD feeding was not associated with reduced expression of inflammatory markers in the ob/ob mice.

Analysis of inflammatory markers [IL-1, IL-6, TNF-α, suppressor of cytokine signaling 3, chemokine (C-C motif) ligand 2 in both liver and adipose tissue, and serine (or cysteine) peptidase inhibitor] in adipose tissues alone revealed no improvements associated with KD feeding (Table 2).

Study 2: Effects of Leptin Administration and Fasting of FGF21 Expression in ob/ob Mice

Leptin treatment induces hepatic FGF21 expression in ob/ob mice.

To establish the possible role of leptin deficiency in mediating increased FGF21 expression seen in chow-fed ob/ob mice, leptin was infused in ob/ob mice and wt controls with free access to chow. Administration of leptin to ob/ob mice led to a 24-fold induction of hepatic FGF21. However, this was associated with a 12% weight loss secondary to a marked reduction in food intake. In leptin-treated wt mice, there was also decreased food intake, but mice did not lose as much weight and failed to induce hepatic FGF21 (Fig. 5, A and B).

Open in a separate window

Fig. 5.

Leptin treatment and fasting induces FGF21 in ob/ob mice. A: weight loss following 72-h administration of saline (filled bars) or leptin (open bars) to wt or ob/ob mice. ^*P < 0.001 leptin treatment vs. saline control. B: induction of FGF21 gene expression resulting from leptin-mediated anorexia. +P < 0.05. C: FGF21 is induced by fasting in ob/ob mice. Filled bars, fed; open bars, fasted. +P < 0.05 and ^*P < 0.001 fasted vs. fed control. D: circulating levels of FGF21 reflect FGF21 gene expression in ob/ob mice fed (filled bars) and fasted (open bars). +P < 0.05 and ^*P < 0.001 fasted vs. fed control.

Fasting for 24 h induces hepatic FGF21 expression in ob/ob mice.

To rule out the possibility that the rise in FGF21 with leptin treatment in ob/ob animals was secondary to decreased food intake as opposed to a direct effect of leptin, we fasted ob/ob mice for 24 h and found that fasting was associated with an additional induction of hepatic FGF21 expression that was reflected in circulating FGF21 levels (Fig. 5, C and D). Combined fasting and leptin treatment did not lead to any additional increase in FGF21 expression in fasted animals (data not shown). In contrast to KD feeding alone, fasting led to a significant increase in circulating ketones in both fasted wt (0.95 ± 0.08 vs. 7.23 ± 0.37 mmol/l) and fasted ob/ob (1.54 ± 0.44 vs. 12.5 ± 2.1 mmol/l) mice.

DISCUSSION

In humans, consumption of KD is associated with weight loss and improvement of metabolic parameters in obese individuals and those with type 2 diabetes (10, 29, 31, 40). We have previously shown that, in wt mice, the rodent version of KD (6), KD reverses DIO and is associated with profound improvements in both circulating lipid profile and glucose tolerance (20). In wt mice with DIO, weight loss is the result of increased energy expenditure in the context of increased leptin levels that normalize as fat mass decreases (20). Weight gain is also prevented by KD in other rodents (7, 35, 42), and it has been suggested that, in juvenile rats, the metabolic effects of KD feeding may be mediated by leptin (37). We therefore undertook an evaluation of the potential effectiveness of KD in a genetic model of obesity and insulin resistance, the ob/ob mouse.

In contrast to DIO animals, KD-fed ob/ob did not lose weight over a 60-day period and maintained the same weight as the chow-fed ob/ob group. We confirmed our previous findings that KD-fed WT mice had an elevated o₂ although notably o₂ was not altered in KD-fed ob/ob mice in either the light or dark periods. Nevertheless, there was a substantial improvement in glucose homeostasis in ob/ob mice eating KD as assessed by a number of parameters. First, in the ad libitum-fed state, ob/ob mice on KD had circulating glucose levels that were 25% lower than ob/ob mice eating chow. Second, KD-fed ob/ob mice had fasting glucose levels that were as low as those seen in wt animals. Third, there was a substantial difference of ~75 and 50%, respectively, in fed and fasted insulin levels between chow- and KD-fed ob/ob groups. The QUCKI metric was markedly improved in KD-fed ob/ob mice, providing a further surrogate marker of increased insulin sensitivity. Dynamic tests of glucose and insulin tolerance were also markedly improved. Of note, glucose excursion during glucose tolerance tests was reduced in KD-fed ob/ob mice, and, although not normalized, there was significant reduction in the area under the glucose curve. This fits with the improvements in insulin sensitivity seen during insulin tolerance testing in which glucose levels fell by 40% in ob/ob mice eating KD compared with <20% in mice eating chow, which is consistent with increased but not completely normalized insulin sensitivity. These results were notable given the absence of weight loss in the ob/ob mice and comparable caloric intake and energy expenditure in fed ob/ob animals. Changes in insulin sensitivity were seen in the absence of any changes in inflammatory markers. However, there were marked differences in serum glucagon levels between chow- and KD-fed ob/ob mouse groups. This contrasts to our previous findings in WT mice, which, as reiterated in this study, have no difference in circulating glucagon levels in chow- and KD-fed states. This decrease in glucagon may contribute significantly to the improved insulin insensitivity seen in the KD-fed ob/ob groups. Whether this is a direct effect of KD on islet glucagon secretion or mediated though another endocrine feedback loop potentially involving FGF21 provides scope for further study.

KD feeding leads to major changes in hepatic gene expression in wt mice with DIO. We found similar changes, including a marked elevation in expression of lipid oxidative and suppression of lipid synthetic genes, in KD-fed ob/ob mice, which occurred in the absence of weight loss. In particular, KD feeding suppressed SCD1 expression in common with other lipid-synthetic genes, including FAS (3, 20). This is notable, since ob/ob mice express hepatic SCD1 at levels over and above those seen in DIO mice (5, 8). The association of improved metabolic status with reduced SCD1 expression may be particularly important given the putative role of SCD1 and products thereof in the regulation of insulin sensitivity and glucose disposal (28, 30, 32, 33). It is possible that KD-mediated downregulation of SCD1 expression is at least in part responsible for improved glucose handling in the ob/ob mice. Expression of SCD1 is regulated by SREBP1c, which also regulates expression of FAS (4, 24, 36), in contrast to wt mice where KD feeding results in reduced SREBP1c expression in both lean and DIO wt mice (20) concomitant with reduced expression of FAS and SCD1. In ob/ob mice, reduction of SREBP1c expression by KD was minimal, perhaps reflecting the high levels of circulating insulin levels.

FGF21 is a metabolic regulator that plays a critical role in mediating the effects of KD in wt animals. In wt mice, we and others have shown that hepatic FGF21 is primarily regulated by PPARα (3, 18). As reported by Lundasen and coworkers (25) we have found that FGF21 expression is increased in the livers of ob/ob mice. In the livers of ob/ob mice, expression of both PPARα and -γ are elevated (11, 27), and it is likely that PPARα is activated by increased circulating endogenous fatty acids in this model. Furthermore, hepatic PPARα target genes CD36, UCP2, and HMGCS2 were robustly elevated in chow-fed ob/ob mouse liver. Because FGF21 potentiates hepatic fatty acid metabolism, basal induction indicates a degree of hormesis in response to endogenous fatty acids. We extended these observations by exposing ob/ob mice to KD, which in wt mice activates PPARα (3). As in wt mice, ob/ob mouse hepatic PPARα target genes, including CD36, ACDL, HADH, UCP2, HMGCS2, and HBDH, were upregulated by KD feeding. In contrast, there was no significant increment in FGF21 beyond the already elevated basal level.

To understand the contribution of leptin deficiency to the increase in hepatic FGF21 expression, we treated ob/ob mice with leptin at doses just above those required to reverse the metabolic phenotype of ob/ob mice (13) and which led to weight loss in wt and ob/ob mice (16). Surprisingly, leptin infusion was associated with a significant additional upregulation of FGF21 in ob/ob mice but not in wt mice. This induction of FGF21 in ob/ob occurs in the context of levels that are already high. The elevation of FGF21 mRNA was observed in the context of leptin-induced hypophagia; this suggests that, in mice with lower food intake, FGF21 rises in a setting where increased β-oxidation of fatty acids from endogenous fuel supplies is required. This is a key role for FGF21 (3, 18). To distinguish between a primary effect of leptin or reduction in food intake, we directly examined the effect of fasting on FGF21 expression and found that, when mice fasted, FGF21 was greatly induced in the ob/ob mice; furthermore, there was no additional increase when mice were fasted and treated with leptin. This indicates that it is reduced food intake, through induction of endogenous ligands such as PPARα, rather than leptin that leads to the increased expression of FGF21 in ob/ob mice with reduced food intake.

In conclusion, we have demonstrated that KD feeding improved the metabolic phenotype of ob/ob mice in the absence of weight loss. Previously, it has been unclear whether beneficial effects of KD feeding on glucose homeostasis were the result of carbohydrate restriction or weight loss. Importantly, in this study with ob/ob mice, we show that effects on glycemia and insulin sensitivity are independent of weight loss. Furthermore, these effects are independent of leptin and occur in the absence of any change in inflammatory markers in either the liver or perigonadal white adipose tissue. It is possible that some of the effects are mediated by altered SCD1 expression, which is known to affect glucose sensitivity and is markedly suppressed by KD feeding. However, effects occur in the absence of any further induction of hepatic FGF21. The elevated basal levels of hepatic FGF21 expression noted in ob/ob mice is likely the result of increased basal PPAR activation. Curiously, KD feeding of ob/ob mice increases expression of a number of PPARα target genes but does not further induce FGF21. However, this effect is not because of maximal expression in the ob/ob model, since reduced caloric intake due to fasting or leptin treatment leads to further induction of FGF21 from baseline. This may be due to lipolysis that occurs under fasting or leptin-treated conditions, yielding increased levels of endogenous PPARα ligands over and above what is generated by KD feeding. Thus further studies are required to fully understand the remarkable effect of KD feeding and its effects on obesity and glucose intolerance.

GRANTS

This work has been supported by generous contributions from the Picower Foundation (E. Maratos-Flier and M. K. Badman), NIDDK Grants DK-069983-01 and DK-56116 (E. Maratos-Flier), and BONRC Pilot and Feasibility Grant DK-46200-15 (M. K. Badman).

ACKNOWLEDGMENTS

REFERENCES